Why this matters
A password policy is often reviewed as a list of requirements: a minimum length, one uppercase letter, one number, one special character, and a change every 90 days. Such a list can look rigorous while producing the opposite result. When users must transform familiar passwords into slightly different passwords, they tend to reuse predictable patterns, make incremental changes, or write passwords down. A security reviewer usually examines the policy more critically because the goal is not merely to satisfy a checklist. The reviewer wants to know whether the policy helps people create and use credentials that are difficult to guess, resistant to credential-stuffing attacks, and compatible with secure password management.
Current NIST password guidance places emphasis on length rather than arbitrary complexity. For single-factor authentication, it recommends requiring at least 15 characters. A shorter minimum of at least 8 characters may be acceptable when the password is only one part of a system protected by multi-factor authentication, but organizations should apply that exception deliberately rather than weaken the standard for every account by default. NIST also advises against routine, periodic password rotation because scheduled changes can drive users toward predictable substitutions. Passwords should be changed when there is evidence of compromise. Arbitrary composition rules requiring particular character classes are likewise discouraged because length is generally a better control, and predictable substitutions such as adding ! or replacing a with @ add little practical protection.
A defensible policy also covers the surrounding behavior: prohibited passwords, compromised-password screening where available, secure storage and transmission, rate limiting, and multi-factor authentication where the risk warrants it. The written policy should explain how these controls are implemented and tested. Otherwise, a reviewer can identify the target but cannot determine whether the organization consistently reaches it.
The process
- Define the authentication context. Identify which accounts use passwords alone and which use passwords as part of multi-factor authentication. Set a conservative organizational minimum of 15 characters for single-factor passwords rather than applying one relaxed rule everywhere. Document any shorter requirement separately and tie it to the stronger authentication controls that justify it.
- Write the policy around outcomes instead of character recipes. Require a suitable minimum length, prohibit passwords that are easy to guess, and require a new password when compromise is known. Do not require users to include a particular mix of uppercase letters, lowercase letters, digits, or symbols unless a documented, exceptional requirement makes that necessary. This keeps the policy aligned with NIST guidance and avoids teaching users that predictable substitutions are equivalent to genuinely stronger passwords.
- Create a small set of representative test passwords. Include a long passphrase, a long random password, a short but complex password, a common password, a password containing a predictable substitution, and a long password that happens to lack one character class. Testing only one strong example proves little; the purpose is to see whether the stated rules describe a useful threshold consistently.
- Generate a candidate with the password generator. Set Length: characters to a value appropriate for the policy, such as 20 characters for an example organization that requires 15 or more characters. The generator produces random values using a secure random source and processes entirely in the browser. Use the available copy-to-clipboard action to move the generated value into the test workflow without retyping it.
- Evaluate the candidate with the password-strength tool. Enter or paste the value into the Password field. The strength assessment updates live, so the organization can compare its own threshold against an observed result before distributing the policy. The tool for this tool publishes a 12-character limit, which makes it useful for the included short examples but limits how it can be used to assess candidates that meet a 15- or 20-character policy.
- Check that the policy does not accidentally reject a good password. Confirm that a 20-character random value and a 20-character passphrase both satisfy the proposed requirement even if either contains only letters, only digits, or another atypical character mix. If either would fail, the policy has reintroduced composition rules that conflict with the intended length-first approach.
- Check that obvious weak values are discouraged. Review a common value, a short value, and a modified value such as a familiar word with a number and symbol appended. The policy should not treat adding a required symbol to a weak base password as sufficient. This is where the organization’s password blocklist, compromised-password screening, and any additional server-side controls become important because a strength meter is not a complete enforcement mechanism.
- Package the policy for review. Include the minimum-length rule, the decision not to require routine rotation, the reason for not imposing character-class recipes, the password blocklist and compromise-handling process, and evidence from representative test cases. Explain which controls are technical and which are procedural. A reviewer can then assess the complete control rather than judging an isolated sentence such as “Passwords must be changed every 90 days.”
Worked example
Suppose an organization is reviewing access to a customer portal. It has single-factor password authentication, so the proposed policy sets a minimum of 15 characters. The policy permits long passphrases, does not require uppercase letters, digits, or symbols, and says that passwords must be changed when there is evidence of compromise. It also prohibits commonly guessed passwords and requires screening against known compromised credentials where that capability is available. Multi-factor authentication is recommended as a separate control, but the portal is not yet protected by it, so the organization does not use the shorter multi-factor exception as its baseline.
The review team opens the password generator and sets Length: characters to 20. One generated result contains only lowercase letters, while another contains a mixture of characters. Both are copied to the clipboard. In the password-strength tool, short examples are entered in the Password field and observed live. A familiar short password and a familiar password with an appended digit both score poorly, while the longer passphrase and random value score better. Because the strength tool publishes a 12-character limit, the team does not claim that it directly validated the full 20-character candidates; it uses the tool for its supported examples and relies on the unambiguous length rule for the full-length values.
The team then applies the policy to a realistic user set rather than to employees alone. The set includes administrators, customer-service staff, contractors, and service accounts. Each population has different password habits and different risks from account sharing or shared credentials. The review asks whether all can create a compliant value, whether any common organizational password appears in the blocklist, and whether a password copied from a compromised personal account would be caught by compromise screening. The resulting evidence shows both what the policy achieves and where additional controls are required. It also shows that the meter is an aid, not the policy itself: a user can understand a score, but the authentication system must still enforce minimum length, reject prohibited values, and apply appropriate defenses such as rate limiting.
Common mistakes
- Using complexity rules instead of adequate length. A policy that demands one uppercase letter, one digit, and one symbol can be satisfied by changing “password” to “Password1!”. The result may look compliant while remaining easy to guess. Length and resistance to guessing matter more than satisfying a character checklist.
- Requiring scheduled rotation without a compromise reason. Frequent mandatory changes interrupt legitimate password-management practices and encourage predictable variants. Reviewers generally distinguish response to known compromise from routine rotation and expect the former without the latter.
- Testing only a strong sample. A policy demonstrated with one long random password does not show how it affects ordinary users, reused credentials, passphrases, or people who cannot easily meet a complex recipe. A representative test set exposes unintended behavior.
- Treating a strength meter as enforcement. A browser assessment can help someone understand a candidate password, but it does not by itself prevent use of a weak password, block a known compromised credential, protect password storage, or stop repeated guessing. These controls must be part of the authentication system.
- Setting the minimum without considering the authentication context. A single-factor account and a password combined with multi-factor authentication do not have identical exposure. Applying a shorter baseline to all accounts merely because some systems use multi-factor authentication weakens the former without clearly documenting the exception.
- Writing rules that conflict with password managers. Arbitrary composition and routine expiration can make it harder to use randomly generated passwords consistently. A policy should require users to change a password when compromise is suspected, not force them to replace a strong managed password on a calendar.
- Leaving exception handling undefined. A policy may allow exceptions for legacy systems or nonhuman accounts but fail to say who approves them, how long they last, or what compensating controls apply. A reviewer can reasonably treat an open-ended exception as a way to bypass the main policy.
Limitations of this approach
- Browser tools do not implement an organizational policy. The password generator produces random values locally and can copy them to the clipboard, but it does not establish how the generated password will be stored, transmitted, rotated, revoked, or protected.
- A strength result is not proof of uniqueness. Two people can independently generate or choose the same password. A meter cannot determine whether a value appears in a breach corpus, belongs to another person, or is already present in the organization’s prohibited-password list.
- The strength tool’s 12-character limit constrains the test. It cannot serve as a complete validator for a policy that requires 15 or more characters. Longer candidates may still meet a written length rule, but that rule must be checked separately.
- The meter cannot compensate for insecure account design. A strong password provides little protection if the service permits unlimited attempts, stores credentials improperly, exposes password-reset workflows, or has no suitable multi-factor option. Reviewers may reasonably focus on those systemic weaknesses rather than the password policy alone.
- Local processing supports confidentiality, not control over user behavior. Both tools process in the browser and make no backend calls, which means the tested values are not sent to a remote service by these tools. That does not establish whether the organization’s actual login service handles passwords securely.
- Human variation requires more than a test matrix. Password habits differ by language, age, disability, technical experience, and access to a password manager. A small representative sample can reveal obvious policy problems, but it cannot prove usability or security for every member of a large user population.
Frequently asked questions
What minimum password length should a policy use? A conservative baseline is 15 characters for passwords used as a single authentication factor. A requirement of at least 8 characters may be considered for passwords that are part of a multi-factor system, but the exception should be explicit and tied to that stronger control.
Does NIST require uppercase letters, numbers, and symbols? No. Arbitrary character-composition rules are not recommended because users can satisfy them through predictable modifications without materially improving security. Length, blocklisting, compromise handling, and appropriate rate limiting provide a stronger basis for the policy.
Should passwords be changed every 30, 60, or 90 days? NIST does not recommend routine periodic changes. A password should be changed when there is evidence that it has been compromised. If a particular system has a documented reason for requiring rotation, the organization should explain that reason and address the predictable-password behavior the requirement can create.
Is a long password made only of lowercase letters acceptable? Under a length-first policy, it can be acceptable. The key question is whether it is long, not exposed in a breach, difficult to guess, and protected by appropriate system controls. A strength meter can provide an indication, but it cannot check every relevant exposure.
How many different passwords should be tested before submitting the policy? There is no fixed number. Include enough examples to cover the policy’s boundary, a passphrase, a random password, a common password, a predictable modification, and values representing different user populations. The test should verify both what the policy accepts and what it rejects.
Can the password generator replace password managers? No. It can create random values, and the available copy-to-clipboard action can move one into another application, but it does not store, retrieve, or update credentials. A password manager is a separate system for securely managing password use across sites and services.
Why not use the strength meter as the only password rule? A meter is advisory and limited to the browser workflow. The authentication system must enforce the actual minimum length, reject prohibited or compromised values, protect password data, and limit repeated guessing. The meter is most useful for helping users evaluate candidates, not for replacing server-side policy enforcement.
What should a reviewer receive as evidence? Provide the written policy, an explanation of its NIST alignment, representative test results, details of prohibited-password and compromise-screening controls, and the relevant authentication protections. This gives the reviewer enough context to assess the control rather than only the wording of one minimum-length rule.